MantisBT - Zandronum
View Issue Details
0001338Zandronum[All Projects] Suggestionpublic2013-05-05 22:262018-09-30 22:27
Edward-san 
Torr Samaho 
normalminoralways
closedfixed 
 
3.03.0 
0001338: Allow clang to compile zandronum (backport)
Since I'm using clang (I believe I'm not alone on this), in order to compile zandronum I need these zdoom changes:

r3589 and r3591 (which fixes an accidental hunk added in r3589). No more, no less.
No tags attached.
child of 0002172closed Torr Samaho Upgrade GZDoom base to 1.8.6 
Issue History
2013-05-05 22:26Edward-sanNew Issue
2014-06-13 17:59WatermelonNote Added: 0009237
2014-06-13 17:59WatermelonStatusnew => feedback
2014-06-13 17:59WatermelonAssigned To => Edward-san
2014-06-13 17:59WatermelonStatusfeedback => assigned
2014-06-13 17:59WatermelonStatusassigned => feedback
2014-06-15 00:14Edward-sanNote Added: 0009358
2014-06-15 00:14Edward-sanStatusfeedback => assigned
2014-06-15 00:52Edward-sanNote Edited: 0009358bug_revision_view_page.php?bugnote_id=9358#r4974
2014-06-15 01:43Edward-sanNote Edited: 0009358bug_revision_view_page.php?bugnote_id=9358#r4975
2015-04-04 21:32ArcoRelationship addedchild of 0002172
2015-04-05 08:16WaTaKiDTarget Version => 3.0
2015-05-09 13:56DuskAssigned ToEdward-san => Torr Samaho
2015-05-09 13:58DuskStatusassigned => needs testing
2015-05-13 07:21Edward-sanNote Added: 0012269
2015-05-13 07:22Edward-sanStatusneeds testing => resolved
2015-05-13 07:22Edward-sanFixed in Version => 3.0
2015-05-13 07:22Edward-sanResolutionopen => fixed
2015-05-17 21:27Edward-sanNote Added: 0012327
2015-05-17 21:34Edward-sanNote Edited: 0012327bug_revision_view_page.php?bugnote_id=12327#r7111
2018-09-30 22:27Blzut3Statusresolved => closed

Notes
(0009237)
Watermelon   
2014-06-13 17:59   
Do we have these in 2.0?
(0009358)
Edward-san   
2014-06-15 00:14   
(edited on: 2014-06-15 01:43)
Nope, the changes are ahead 2.5.0.

[edit] uh ... I don't agree with the message added by the first commit:


// [CE] Clang doesn't yet support some inline ASM operations so I disabled it for that instance


The problem is that Clang is perfectly right to reject that assembly code, and gcc is wrong to allow that. I asked some asm experts if there was a workaround, they told me it's not a good idea to use asm like that at all. :\

[edit2] It seems there's a workaround:


-: "%3", "%4", "st(1)", "st(7)" );
+: "st(1)", "st(2)", "st(3)", "st(4)", "st(5)", "st(6)", "st(7)" );


(0012269)
Edward-san   
2015-05-13 07:21   
Apart from the compiler error I pointed out in ticket 0002172, with 3.0 clang compiles fine ( and also gcc5, which had the same compiler errors as clang for 'dumb' code ).
(0012327)
Edward-san   
2015-05-17 21:27   
(edited on: 2015-05-17 21:34)
I noticed that clang still has an error when compiling src/p_spec.h . This is fixed by zdoom in'https://github.com/rheit/zdoom/commit/310979e9e6d64a54bf0b7a0df4627f31f5c06f6c [^]' .


edit: nvm, that change is present in 2.7.1, so this can wait..